A New Model of Trapping and Relative

Permeability Hysteresis for All

Wettability Characteristics
Elizabeth J. Spiteri, SPE, Chevron Energy Technology Company; Ruben Juanes, SPE, Massachusetts Institute of
Technology; Martin J. Blunt, SPE, Imperial College London; and Franklin M. Orr, Jr., SPE, Stanford University
Summary
The complex physics of multiphase flow in porous media are
usually modeled at the field scale using Darcy-type formulations.
The key descriptors of such models are the relative permeabilities
to each of the flowing phases. It is well known that, whenever the
fluid saturations undergo a cyclic process, relative permeabilities
display hysteresis effects.
In this paper, we investigate hysteresis in the relative perme-
ability of the hydrocarbon phase in a two-phase system. We pro-
pose a new model of trapping and waterflood relative permeability,
which is applicable for the entire range of rock wettability condi-
tions. The proposed formulation overcomes some of the limitations
of existing trapping and relative permeability models. The new
model is validated by means of pore-network simulation of pri-
mary drainage and waterflooding. We study the dependence of
trapped (residual) hydrocarbon saturation and waterflood relative
permeability on several fluid/rock properties, most notably the
wettability and the initial water saturation. The new model is able
to capture two key features of the observed behavior: (1) non-
monotonicity of the initial-residual curves, which implies that wa-
terflood relative permeabilities cross; and (2) convexity of the
waterflood relative permeability curves for oil-wet media caused
by layer flow of oil.
Introduction
Hysteresis refers to irreversibility or path dependence. In mul-
tiphase flow, it manifests itself through the dependence of relative
permeabilities and capillary pressures on the saturation path and
saturation history. From the point of view of pore-scale processes,
hysteresis has at least two sources: contact angle hysteresis, and
trapping of the nonwetting phase.
The first step in characterizing relative permeability hysteresis
is the ability to capture the amount of oil that is trapped during any
displacement sequence. Indeed, a trapping model is the crux of any
hysteresis model: it determines the endpoint saturation of the hy-
drocarbon relative permeability curve during waterflooding.
Extensive experimental and theoretical work has focused on the
mechanisms that control trapping during multiphase flow in po-
rous media (Geffen et al. 1951; Lenormand et al. 1983; Chatzis
et al. 1983). Of particular interest to us is the influence of wetta-
bility on the residual hydrocarbon saturation. Early experiments in
uniformly wetted systems suggested that waterflood efficiency de-
creases with increasing oil-wet characteristics (Donaldson et al.
1969; Owens and Archer 1971). These experiments were per-
formed on cores whose wettability was altered artificially, and the
results need to be interpreted carefully for two reasons: (1) reser-
voirs do not have uniform wettability, and the fraction of oil-wet
pores is a function of the topology of the porous medium and
initial water saturation (Kovscek et al. 1993); and (2) the coreflood
experiments were not performed for a long enough time, and not
enough pore volumes were injected to drain the remaining oil
layers to achieve ultimate residual oil saturation. In other coreflood
experiments, in which many pore volumes were injected, the ob-
served trapped/residual saturation did not follow a monotonic
trend as a function of wettability, and was actually lowest for
intermediate-wet to oil-wet rocks (Kennedy et al. 1955; Moore and
Slobod 1956; Amott 1959). Jadhunandan and Morrow (1995) per-
formed a comprehensive experimental study of the effects of wet-
tability on waterflood recovery, showing that maximum oil recov-
ery was achieved at intermediate-wet conditions.
An empirical trapping model typically relates the trapped (re-
sidual) hydrocarbon saturation to the maximum hydrocarbon satu-
ration; that is, the hydrocarbon saturation at flow reversal. In the
context of waterflooding, a trapping model defines the ultimate
residual oil saturation as a function of the initial water saturation.
The most widely used trapping model is that of Land (1968). It is
a single-parameter model, and constitutes the basis for a number of
relative permeability hysteresis models. Other trapping models are
those of Jerauld (1997a) and Carlson (1981). These models are
suitable for their specific applications but, as we show in this
paper, they have limited applicability to intermediate-wet and oil-
wet media.
Land (1968) pioneered the definition of a flowing saturation,
and proposed to estimate the imbibition relative permeability at a
given actual saturation as the drainage relative permeability evalu-
ated at a modeled flowing saturation. Lands imbibition model
(1968) gives accurate predictions for water-wet media (Land
1971), but fails to capture essential trends when the porous me-
dium is weakly or strongly wetting to oil. The two-phase hysteresis
models that are typically used in reservoir simulators are those by
Carlson (1981) and Killough (1976). A three-phase hysteresis
model that accounts for essential physics during cyclic flooding
was proposed by Larsen and Skauge (1998). These models have
been evaluated in terms of their ability to reproduce experimental
data (Element et al. 2003; Spiteri and Juanes 2006), and their
impact in reservoir simulation of water-alternating-gas injection
(Spiteri and Juanes 2006; Kossack 2000). Other models are those
by Lenhard and Parker (1987), Jerauld (1997a), and Blunt (2000).
More recently, hysteresis models have been proposed specifically
for porous media of mixed wettability (Lenhard and Oostrom
1998; Moulu et al. 1999; Egermann et al. 2000).
All of the hysteresis models described require a bounding
drainage curve and either a waterflood curve as input, or a calcu-
lated waterflood curve using Lands model. The task of experi-
mentally determining the bounding waterflood curves from core
samples is arduous, and the development of an empirical model
that is applicable to non-water-wet media is desirable. In this pa-
per, we introduce a relative permeability hysteresis model that
does not require a bounding waterflood curve, and whose param-
eters may be correlated to rock properties such as wettability and
pore structure.
Because it is difficult to probe the full range of relative per-
meability hysteresis for different wettabilities experimentally, we
use a numerical toolpore-scale modelingto predict the trends
in residual saturation and relative permeability. As we discuss
later, pore-scale modeling is currently able to predict recoveries
and relative permeabilities for media of different wettability reli-
ably (Dixit et al. 1999; ren and Bakke 2003; Jackson et al. 2003;
Valvatne and Blunt 2004; Al-Futaisi and Patzek 2003, 2004). We
Copyright 2008 Society of Petroleum Engineers
This paper (SPE 96448) was first presented at the 2005 SPE Annual Technical Conference
and Exhibition, Dallas, 912 October, and revised for publication. Original manuscript re-
ceived for review 21 July 2005. Revised manuscript received 11 April 2007. Paper peer
approved 23 January 2008.
277 September 2008 SPE Journal
will use these predictions as a starting point to explore the behavior
beyond the range probed experimentally.
In summary, this paper presents a new model of trapping and
waterflood relative permeability, which is able to capture the be-
havior predicted by pore-network simulations for the entire range
of wettability conditions.
Pore-Scale Modeling of Trapping and Hysteresis
In pore-network modeling, the pore space is described by a net-
work of pores connected by throats with an idealized geometry. A
set of physically based rules describe the configuration of the
fluids within each pore and throat, as well as the mechanisms for
the displacement of one fluid by another. This approach was
pioneered by Fatt (1956) and has received increasing attention
over the past decade. Blunt (2001) and Blunt et al. (2003) provide
a detailed description of the fundamentals and applications of
pore-network modeling, together with an extensive literature re-
view. One of the successful application areas of pore-network
models is the prediction of multiphase flow properties, such as
capillary pressure and relative permeability. This success hinges
on the following:
1. The ability to reproduce the essential geometric features of
the pore space of real rocks. A realistic 3D pore-space character-
ization may be obtained in a variety of ways: assembly of 2D
sections to form a 3D image (Holt et al. 1996); direct X-ray mi-
crotomography of the 3D pore space (Spanne et al. 1994); sto-
chastic 3D modeling with statistics inferred from 2D thin sections
(Ioannidis and Chatzis 2000); and process-based reconstruction in
which grain deposition, compaction, and cementation are modeled
(Bryant and Blunt 1992; ren and Bakke 2002).
2. The ability to capture wettability effects. Most pore-network
models used today introduce wettability effects based on the pore-
level scenario of wetting proposed by Kovscek et al. (1993). Their
model mimics the saturation change typical of a hydrocarbon res-
ervoir. The medium is initially filled with water, and the rock
surfaces are water-wet. During oil migration, the oil invades the
pore space, altering the wettability of the solid surface in contact
with the oil. In this fashion, the network displays mixed wettabil-
ity: a fraction of a pore or throat may be oil-wet, while the corners
and crevices not in contact with the oil remain water-wet.
The combination of realistic pore geometry/topology and cor-
rect characterization of displacement and trapping mechanisms has
allowed pore-network models to predict hysteretic capillary pres-
sure and relative permeability curves under a wide range of wet-
tability characteristics.
Experimental measurements of hysteretic relative permeability
for mixed-wet and oil-wet media are scarce (Jadhunandan and
Morrow 1995; Oak 1991). Given the success of pore-network
models to reproduce experimental data (ren and Bakke 2003;
Jackson et al. 2003; Valvatne and Blunt 2004), in this work we
have adopted the use of pore-network modeling as a way to in-
vestigate the full spectrum of wettability conditions. Pore-network
simulations results are taken as data to develop and validate
empirical trapping and hysteresis models.
Description of the Pore-Network Simulations. We used the two-
phase flow pore-network simulator developed by Valvatne and
Blunt (2004). The model has similarities with other network mod-
els (Al-Futaisi and Patzek 2003; ren et al. 1998; Patzek 2001). A
full description of the model is given in the previous references.
We used a 3D pore-network of a Berea sandstone developed by
Bakke and ren (1997). The model is a cube of 333 mm
3
containing 12,349 pores and 26,146 throats. The absolute perme-
ability of the rock is 2287 md and the net porosity is 0.183, where
0.0583 is clay-bound or microporosity.
We investigated sequences of two consecutive displacements:
primary drainage (oil invasion) and waterflooding. During primary
oil drainage the network, which is initially filled with water, is
assumed to be strongly water-wet with a receding contact angle

r
0. As the oil invades the largest pores first in piston-like
displacement, the water recedes and is squeezed to the crevices and
pore throats until a very high capillary pressure or a target oil
saturation is reached. At this point, the surface of the rock in
contact with oil will undergo wettability alteration, while the cor-
ners and elements that still contain only water remain strongly
water-wet. Wettability alteration is accounted for by changing the
contact angle. In principle, one could change the advancing contact
angle
a
and the receding contact angle
r
independently. In this
work, however, we used a correlation proposed by Morrow (1975)
to link both the advancing and receding contact angles with an
intrinsic contact angle
i
. This relationship is shown in Fig. 1.
During waterflooding, there are several physical mechanisms
by which the water can displace the oil in place (Lenormand et al.
1983). These mechanisms include piston-type displacement, co-
operative pore-body filling, and snap-off. The predominance of
any given displacement mechanism is strongly dependent on the
wettability (specified by the advancing contact angle). These dis-
placement processes and their implementation are described in
detail in the literature (Valvatne and Blunt 2004; Al-Futaisi and
Patzek 2003; ren et al. 1998; Patzek 2001). After individual
displacement events, the transport properties are calculated. The
equations for absolute permeability, relative permeability, and
other transport parameters can also be found in the literature (Val-
vatne and Blunt 2004; ren et al. 1998; Patzek 2001; Blunt and
King 1991). Following this procedure, pore-network simulations
have been shown to reproduce experimental capillary pressure and
relative permeability curves both in primary drainage and water-
flooding (ren et al. 1998) and for a variety of wettability condi-
tions (ren and Bakke 2003; Jackson et al. 2003; Valvatne and
Blunt 2004).
We investigated the full range of wetting conditions after wet-
tability alteration, by choosing average intrinsic contact angles
between 20 and 160. Because of pore-scale inhomogeneities in
the rock minerals and surface roughness, assigning a uniform con-
tact angle throughout is unrealistic. Therefore, we assigned contact
angles throughout the network randomly within 20 of the aver-
age value.
In this work, we assume that the contact angle distribution in
originally oil-filled pores is independent of S
oi
. In reality, a higher
S
oi
represents a larger initial capillary pressure that may cause
protective water layers to collapse in oil-filled pores, causing a
more significant wettability alteration than for low S
oi
(Kovscek
et al. 1993). However, this simple characterization of wettability
has been shown to be sufficient to predict trends in recovery seen
experimentally (Valvatne and Blunt 2004). Also, we consider sys-
tems in which there is a relatively uniform distribution of wetta-
bilitywe do not consider mixed-wet media where initially oil-
filled pores may be water-wet or oil-wet with two distinct distri-
butions of contact angle in the same rock.
For each contact angle distribution we performed a series of
displacement pairs (oil invasion and waterflooding), with a differ-
Fig. 1Relationship between receding and advancing contact
angles on a rough surface, as a function of intrinsic contact
angle measured at rest on a smooth surface (Morrow 1975) (Fig.
from Valvatne and Blunt 2004).
278 September 2008 SPE Journal
ent target oil saturation S
oi
ranging from a very small value (al-
most no oil invasion) to the maximum value possible (connate
water conditions). Next, we present results from an extensive set
of simulations for different contact angles and different target
oil saturations.
Pore-Network Predictions of Trapping. The use of a pore-
network simulator allows us to quantify the effects of wettability
and initial oil saturation on the trapping of oil during waterflood-
ing. The main results of the pore-network simulations are compiled
in Fig. 2, where we plot the initial-residual (IR) curves for differ-
ent intrinsic contact angles. For a given curve (i.e., for a specific
value of the intrinsic contact angle after wettability alteration),
each point denotes the trapped oil saturation S
ot
that corresponds to
a particular initial oil saturation S
oi
. The initial oil saturation is
dictated by the point at which primary drainage ceases (and wa-
terflood starts), and the trapped oil saturation is the value of un-
recoverable (or residual) oil after waterflooding.
The most noteworthy characteristic of the IR curves shown in
Fig. 2 is that they do not display a monotonically increasing be-
havior for mixed-wet and oil-wet media. This means that, for
oil-wet media, higher oil saturation after the initial oil invasion
may lead to lower residual oil saturation after waterflooding. This
counterintuitive behavior is analyzed and explained next, in terms
of pore-scale fluid configurations and displacement mechanisms.
For water-wet media (small contact angles), trapping during
waterflooding is controlled by snap-off. As the initial oil saturation
is increased, oil is pushed into smaller and smaller pores. During
waterflooding, water fills the smallest pores first and snaps off
more and more oil. The amount of trapped oil increases mono-
tonically with increasing initial oil saturation simply because there
is more oil to trap. As the contact angle increases, there is a
crossover from trapping by snap-off to trapping by bypassing, as
the water tends to advance in a connected front with piston-like
advance on throats and cooperative pore-filling. We begin to see a
non-monotonic behavior because, as the initial water saturation
increases, we have more water filling from the pores and throats in
connected patches from throats that are initially water-filled. When
these patches join up, there can be trapping, as oil becomes
stranded between these clusters. Low initial water saturation (i.e.,
high initial oil saturation) implies that there are few water clusters
and little chance for bypassing. Trapping increases as the number
of clusters increases and then declines again as there is less oil to
trap in the first place. For intrinsic contact angles above 90 (ad-
vancing contact angles greater than 120), we observe the same
behavior, but with even less trapping. The reason is the presence of
oil layers within the network. Water fills the largest pores and
throats in an invasion percolation-like process. Oil remains con-
nected in layers sandwiched between water in the center of an
element and water in the corners. These layers lead to very little
trapping, although the oil relative permeability is very low. These
layers are stable until the two water/oil interfaces meet. High initial
water saturation means that water bulges out in the corners causing
these interfaces to meet, trapping more oil than for low initial
water saturation. Again, for sufficiently large S
wi
, (low S
oi
), there
is less trapping simply because there is less oil to trap.
We should also mention that the extremely low trapped oil
saturations for very high initial oil saturations are an artifact of the
criterion used for ascertaining the stability of oil layers (Valvatne
and Blunt 2004). In the future, we plan to incorporate a stability
criterion based on free-energy balance that predicts that oil layers
become unstable before the two water/oil interfaces touch each
other (van Dijke et al. 2004). For practical purposes, however,
such high initial oil saturations are never achieved during migra-
tion of oil into realistic reservoirsthe initial water saturation is
typically much higher than 5%.
In conclusion, the trapping mechanisms that we have indicated
allow for a physical explanation of the non-monotonic behavior of
the initial-residual curves. A complete picture of the trapping re-
lation is given in Fig. 3 as a trapping surface; that is, a surface that
describes the residual oil saturation as a function of the initial oil
saturation and the intrinsic contact angle after wettability alteration.
The trapped oil saturation dictates the endpoint of the relative
permeability waterflood curves. An important practical conse-
quence of the nonmonotonic relation of trapped vs. initial oil satu-
ration for mixed- to oil-wet media is that waterflood scanning
curves will cross, as sketched in Fig. 4.
Neither of these featuresnonmonotonic trapping relation and
crossover of waterflood relative permeability curvesare present
in existing empirical models. This motivates the development of
new empirical trapping and hysteresis models that reproduce the
observed behavior.
Pore-Network Predictions of Waterflood Relative Permeabil-
ity. It is important to understand the trapping mechanisms that
ultimately define the shape of the relative permeability curves
during waterflooding. The trapping model determines the endpoint
residual saturations when the capillary pressure is lowered to an
extremely low value. The trapping mechanisms at this point should
Fig. 2Trapped oil saturation (S
ot
) as a function of the initial oil
saturation (S
oi
) for different intrinsic contact angles (). Note
that the curves are nonmonotonic except for strongly water-
wet media.
Fig. 3Contour plot of the trapping surface. Trapped oil satu-
ration depends on the initial oil saturation (S
oi
) and rock wetta-
bility in terms of the intrinsic contact angle ().
279 September 2008 SPE Journal
not be generalized for the entire waterflood process. Because dif-
ferent competing trapping mechanisms may dominate at different
capillary pressure levels, this affects the shape of the relative per-
meability curve.
In water-wet systems, we have already mentioned that trapping
is primarily due to snap-off. After a certain point during water-
flooding, the flowing oil phase becomes trapped because of the
invading water phase. However, this mechanism does not operate
right away. At the beginning of the waterflood process, the non-
wetting-phase relative permeability is slightly higher than the
drainage relative permeability. This is seen in the two-phase ex-
periments performed by Oak (1990) in water-wet Berea sandstone
(see Fig. 5).
During primary drainage, the oil preferentially fills large pores,
leaving water residing in narrower throats and in the corners of the
pore space. At the end of primary drainage, many oil-filled pores
have only a single connecting throat that is also oil-filled. These
dead-end pores may contain a large saturation but do not contrib-
ute to the connectivity of the oil. During waterflooding, pore-
filling is favored in pores that have many surrounding water-filled
throats, essentially these dead-end pores (Lenormand et al. 1983).
Thus initially there is a cascade of pore-filling, where the oil
saturation decreases with little decrease in oil relative permeabil-
ity. This process competes with snap-off, which traps oil and in
contrast leads to a large decrease in relative permeability. For
water-wet media, the former process generally is more significant
at high oil saturation, giving the typical hysteresis patterns seen in
Fig. 5. The pore-network model is able to reproduce this behavior
with a quantitative agreement with experiment (Valvatne and
Blunt 2004); see Fig. 6.
In contrast, the trapping mechanisms that control the shape of
the oil waterflood relative permeabilities in oil-wet media are very
different from those of a water-wet rock. At the beginning of the
waterflood, water percolates through the largest pores, leading to a
significant reduction in the oil relative permeability. The oil, which
remains connected through oil layers, drains down to very low
saturations but at a low rate because of the small conductance of
these layers. The shape of these relative permeability curves, as
predicted by pore-network simulations, is shown in Fig. 7.
Development and Validation of a New Model of
Relative Permeability Hysteresis
In the previous section, we used pore-network modeling to high-
light the following features of wettability effects on the waterflood
relative permeability:
Fig. 5Oil relative permeability curves for an water/oil sys-
tem from Oaks (1990) experimental data in water-wet
Berea sandstone.
Fig. 6Oil relative permeabilities generated from pore-network
simulation in strongly water-wet media,
i
[0,20].
Fig. 7Oil relative permeabilities generated from pore-network
simulation in strongly oil-wet media,
i
[160,180].
Fig. 4Waterflood relative permeability curves that cross be-
cause of the nonmonotonic trapping relationship.
280 September 2008 SPE Journal
1. The IR curves are not monotonic for media that are not
strongly water-wet (Fig. 2).
2. For intermediate-wet and oil-wet media, the scanning curves
of oil relative permeability may cross (Fig. 4).
3. In strongly water-wet media, the trapped oil saturation is
high, but the waterflood relative permeability may be higher than
the drainage relative permeability at high oil saturations (Fig. 6).
4. In contrast, for strongly oil-wet media, the trapped oil satu-
ration is low, but the waterflood relative permeability decreases
sharply at high oil saturations (Fig. 7).
Clearly, this markedly different behavior in water-wet and oil-
wet media needs to be incorporated in the empirical model. We
start by describing a new trapping submodel, and we follow with
the proposed waterflood relative permeability model.
The Trapping Model. We begin by reviewing some of the exist-
ing trapping models. These models were originally designed to
account for gas trapping, but for consistency we will treat them for
oil trapping. We then formulate a new model and assess its per-
formance for the full spectrum of wettability conditions.
Land Trapping Model. The Land model (1968) is the most
widely used empirical trapping model. Most relative permeability
models that incorporate hysteresis (Jerauld 1997a; Killough 1976;
Larsen and Skauge 1998; Lenhard and Parker 1987; Blunt 2000;
Lenhard and Oostrom 1998) are based on it. It was developed to
predict trapped gas saturation as a function of the initial gas satu-
ration based on published experimental data from water-wet sand-
stone cores (Holmgren and Morse 1951; Kyte et al. 1956; Darda-
ganian 1957).
The trapped nonwetting phase saturation is computed as:
S
ot
S
oi
=
S
oi
1 + CS
oi
, . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (1)
where S
oi
is the initial oil saturation, or the saturation at the flow
reversal, and C is the Land trapping coefficient.
The Land coefficient is computed from the bounding oil inva-
sion and waterflood curves as follows:
C =
1
S
ot,max

1
S
o,max
, . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (2)
where S
o,max
is the maximum oil saturation, and S
ot,max
is the
maximum trapped oil saturation, associated with the bounding
waterflood curve. All these quantities are illustrated in Fig. 8. The
value of the Land trapping parameter is dependent on the type of
rock and fluids (Spiteri and Juanes 2006).
Carlson Trapping Model. A simplified hysteresis model pro-
posed by Carlson (1981) implicitly defines a trapping model. The
Carlson model requires the bounding drainage and waterflood
curves. The trapped oil saturation is determined by shifting the
bounding waterflood curve to intersect the intermediate initial oil
saturation at the flow reversal. The idea behind Carlsons inter-
pretation is to use the model of the waterflood relative permeabil-
ity scanning curves as being parallel to each other. This geometric
extrapolation procedure is illustrated in Fig. 9.
The trapped nonwetting-phase saturation is computed as
S
ot
= S
ot,max
S
o
, . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (3)
where S
o
is the shift in the waterflood scanning curve with re-
spect to the imbibition bounding curve (see Fig. 9).
This model is adequate if the intermediate scanning curves are
almost parallel and there is little curvature in the waterflood curve.
The model is problematic when the system is oil-wet. The large
curvature of the bounding waterflood relative permeability curve
at low saturations does not allow prediction of intermediate rela-
tive permeability curves, because any shifting will make the end-
point trapped gas saturation negative, a nonphysical value.
Jerauld Trapping Model. Jeraulds trapping model (1997a) is
an extension of the Land trapping model that accounts for the
plateau observed in the IR curves for mixed-wet rocks (1997b).
The trapped nonwetting-phase saturation is given by:
S
ot
=
S
oi
1 + CS
oi
1+bC
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (4)
In the original publication (Jerauld 1997a), the expression of the
trapped saturation S
ot
was given in terms of the residual saturation
achieved when S
oi
1, S
or
1/(C+1). Jerauld introduced a second
tuning parameter b in addition to the Land coefficient. If this
parameter is set to zero, Jeraulds model reduces to the Land
trapping model. When this parameter is equal to one, the trapping
curve has a zero slope at S
oi
1. Although Jerauld argued that the
IR curves should not have a negative slope, his model allows for
such behavior if b1.
Although the fit of Jeraulds model to the pore-network data
was good for water-wet and intermediate-wet conditions (for
which the model was designed), it was not as satisfactory for
strongly oil-wet media. One of the reasons is that Jeraulds model
assumes that the IR curve has a unit slope near the origin. This
behavior does not conform to pore-network predictions (see the
curve corresponding to 140 in Fig. 2).
Fig. 8Parameters required in the evaluation and application of
the Land trapping model.
Fig. 9Geometric extrapolation of the oil relative permeability
and trapped saturation during waterflooding, as proposed by
Carlson (1981).
281 September 2008 SPE Journal
A New Trapping Model. We notice that the shapes of the
trapping curves (Fig. 2) may be fit to a parabola. We establish the
following simple quadratic relationship between the trapped oil
saturation S
ot
and the initial oil saturation S
oi
:
S
ot
= S
oi
S
oi
2
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (5)
The parameters and correspond to the initial slope and the
curvature of this curve, respectively. These parameters were tuned
to minimize the least squared error between the model prediction
and the pore-network simulation data. The optimization is con-
strained by the following restrictions:
0 1, 0. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (6)
For an initial slope greater than 1, the trapping model would pre-
dict more trapped oil than what was originally present, which is not
physically possible. The optimal parameters and are shown
in Fig. 10 as functions of the intrinsic contact angle. The proposed
trapping model has two parameters and, therefore, we expect a
better fit to the pore network simulation results than the Land
trapping model. The presence of an additional parameter is entirely
justified, however, by the need to capture the non-monotonic be-
havior of the IR curves.
The performance of the optimization is illustrated in Fig. 11.
Notice that for water-wet media, the model tends to slightly over-
estimate the trapped oil saturation when the initial oil saturation is
high. This is because of the constraint in the optimization, and the
inability of the model to achieve the desired curvature at the de-
sired location. The important consideration is that the trapping
model reproduces the observed trapping behavior for all wettabil-
ity conditions, even if it may slightly overestimate the trapping of
the bounding waterflood curve (S
oi
S
o,max
1S
wc
). We should
also mention that the pore-network simulator is likely to underes-
timate the trapping for the bounding curves because of an overly
optimistic criterion for the stability of oil layers. If a new, free-
energy based stability criterion is implemented (van Dijke et al.
2004), we expect a better agreement between the trapping model
and the pore-network predictions.
When the parameters calculated from the optimization are em-
ployed, the resulting trapping surface is shown in Fig. 12. This
surface should be compared with the one obtained from pore-
network simulations (Fig. 3).
The Waterflood Relative Permeability Model. Most existing
relative permeability hysteresis models (Jerauld 1997a; Killough
1976; Larsen and Skauge 1998; Lenhard and Parker 1987; Blunt
2000) either require a bounding waterflood curve or model this
curve according to Lands (1968) waterflood relative permeability
model. The development of his model is described next. The new
relative permeability model we propose is an extension of Lands
model to account for the different pore occupancies at different
wettability conditions.
Land Waterflood Model. As a prelude to the development of
the new waterflood relative permeability model proposed in this
work, we revisit the derivation of Lands relative permeability
model. The basis of Lands formulation is to express the water-
flood relative permeability k
ro
i
at a given oil saturation (S
o
) as being
equal to the drainage permeability k
ro
d
evaluated at a flowing oil
saturation S
of
(see Fig. 13):
k
ro
i
S
o
= k
ro
d
S
of
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (7)
At any bulk saturation S
o
, the flowing oil saturation S
of
and the
trapped saturation S
o
are related by
S
o
= S
o
S
of
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (8)
Land makes the assumption that the trapped saturation S
o
is the
cumulative trapped saturation at a given point in the waterflood
process and that this quantity increases as more of the flowing
saturation becomes trapped. He assumes that the maximum
amount of cumulative trapping, equal to the trapped saturation
determined by his trapping model (Eq. 1) occurs when the flowing
saturation becomes zero [S
o
=S
ot
(S
oi
)]. It is important to note that
in Lands formulation, it is necessary to obtain the maximum
trapped oil saturation S
ot,max
from a coreflood experiment in order
to extract the appropriate Land trapping coefficient C.
The intermediate trapped saturation S
o
is equal to the cumu-
lative trapped saturation S
ot
minus the amount of oil that is still
flowing and will eventually be trapped:
S
o
= S
ot
S
ot
S
of
, . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (9)
where
S
ot
S
ot
S
oi
=
S
oi
1 + CS
oi
, . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (10)
and
S
ot
S
of
=
S
of
1 + CS
of
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (11)
Substituting Eqs. 9 through 11 in Eq. 8, one obtains
S
of
2
S
o
S
ot
S
of

1
C
S
o
S
ot
= 0. . . . . . . . . . . . . . . . . . . . . (12)
Solving this quadratic equation for S
of
and taking the positive root:
S
of
=
1
2
S
o
S
ot
+
S
o
S
ot

2
+
4
C
S
o
S
ot
. . . . . . . . (13)
Although Lands assumptions are generally valid for water-wet
media, they do not hold for oil-wet media. In water-wet media, we
noticed that the experimental and pore-network waterflood curves
are sometimes above the drainage curves for high oil saturations
(Figs. 5 and 6). Lands model will generally underestimate the
relative permeabilities in this region under the assumption that the
hydrocarbon phase will immediately be subjected to snap-off. In
oil-wet media, the ultimate residual saturations are often very low
because of oil layer drainage. Initially, there is a sharp decrease in
the waterflood relative permeability (Fig. 7). However, at low oil
saturations, oil layer drainage is the dominant mechanism, which
leads to low oil residual saturations achieved at very low oil rela-
tive permeabilities.
In Fig. 14, we compare Lands waterflood relative permeability
model to the pore-network simulation results. The thick red line is
the drainage relative permeability curve, obtained from the pore
network simulator. The blue circles correspond to the pore network
imbibition curve and the thin blue line to the Land waterflood
relative permeability model. The Land trapping model predicts the
experimental data fairly well for water-wet media, but is unable to
capture the convex shape of the waterflood curve characteristic of
Fig. 10Dependence of parameters and of the proposed
quadratic trapping model on the intrinsic contact angle of ini-
tially oil-filled pores for a network model of a Berea sandstone.
282 September 2008 SPE Journal
Fig. 12Trapping surface determined with the new trapping model.
Fig. 13Trapped and flowing saturations used to determine
the waterflood relative permeability from the drainage rela-
tive permeability.
Fig. 11Performance of the new trapping model: IR curves calibrated against pore-network simulation data.
283 September 2008 SPE Journal
oil-wet conditions. Indeed, for intrinsic contact angles greater than
80, Lands model predicts reversible relative permeability
curveswaterflood relative permeability curves coincide with the
primary drainage curves.
A New Waterflood Model. Lands waterflood model hinges on
the assumption that the trapped saturation increases monotonically
during waterflooding. This assumption does not allow reproduc-
tion of the convex shape of the observed relative permeability
curves in intermediate-wet and oil-wet media. We modify Eq. 9
as follows:
S
o
= S
ot
S
oi
S
ot
S
of
S
o
S
ot
S
oi
S
o
S
oi
. . . . . . . . . (14)
The last term in this equation is designed to capture the convexity
of the waterflood curves in oil-wet media. It satisfies the following
essential requirements: (1) the flowing saturation S
of
equals zero
when the bulk saturation reaches the ultimate trapped saturation
S
ot
(S
oi
); and (2) the flowing saturation is equal to the bulk satura-
tion (S
of
=S
o
) at the beginning of the waterflood (S
o
=S
oi
) and no oil
has yet been trapped. The parameter is an additional parameter
of the formulation, which should depend on rock type and wetta-
bility characteristics.
Substituting Eq. 14 into Eq. 8, we obtain
S
of
= S
o
S
ot
S
oi
+ S
ot
S
of
+ S
o
S
ot
S
oi
S
o
S
oi
, . . . . . (15)
where the trapped saturation is given by the new trapping model:
S
ot
S
oi
= S
oi
S
oi
2
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (16)
We substitute Eq. 16 in Eq. 15 and solve for S
of
, to obtain a new
model for the flowing oil saturation:
S
of
=
1
2
a 1 + 1
2
+ 4S
o
S
ot
+ S
o
S
ot
S
o
S
oi
.
. . . . . . . . . . . . . . . . . . . . . . . . . . (17)
This expression of the flowing saturation is then used in Eq. 7 to
evaluate the waterflood relative permeability.
The tuning parameter can be obtained from fitting the model to
experimental bounding waterflood curves. In this investigation, we
used bounding waterflood curves obtained from pore-network
simulation. The dependence of the parameter on the intrinsic con-
tact angle is shown in Fig. 15. The trends in this relationship are
the ones expected. For water-wet media, the parameter is negative,
indicating that the Land trapping model overestimates the trapped
saturation and subsequently underestimates the relative permeabil-
ity. For oil-wet media, this parameter takes positive values, which
allows for the model waterflood curve to be below the one pre-
dicted by the Land model.
The performance of the combined trapping and waterflood
models is shown in Fig. 16. Unlike Lands trapping model, the
new model provides a suitable fit to the bounding waterflood
curves determined by pore-network simulations for all contact
angles. The dark circle represents the trapped saturation deter-
mined from the new trapping model. The trapped saturations pre-
dicted by the model do not always match the experimental end-
points. This is why the model relative permeabilities do not fully
agree with the pore-network results, especially for saturations
close to the ultimate residual saturation.
In Fig. 17, we compare the new waterflood model with
pore-network simulated data for a set of intermediate scan-
Fig. 14Comparison of Lands trapping model for the bounding relative permeability curves with pore-network simulation data.
284 September 2008 SPE Journal
ning curves. Model predictions were obtained using the same pa-
rameters determined from the trapping curves and the bounding
waterflood curves.
Wettability Correlations. In this paper, we have consistently used
the intrinsic contact angle
i
as a measure of wettability. This
parameter is almost impossible to determine with any certainty in
the laboratory because most rocks are characterized by a large
range of contact angles. Moreover, we have used a particular
model (Morrow 1975) that links the intrinsic contact angle with the
receding and advancing contact angles.
Ideally, one would correlate the trapping parameters and
and the waterflood parameter with a measure of the overall
wettability characteristics of the rock that can be determined in the
lab. In fact, previous investigations (ren and Bakke 2003;
Valvatne and Blunt 2004) have shown that pore-network models
are able to perform quantitative predictions of laboratory wetta-
bility measurements.
Common measures of wettability are the Amott wettability in-
dices I
w
and I
o
(1959). A strongly water-wet medium is associated
with I
w
1 and I
o
0, while values of I
w
0 and I
o
1 correspond
to a strongly oil-wet medium. The Amott-Harvey index I
wo
is
probably the most popular measure of wettability and is defined as
I
wo
= I
w
I
o
, . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (18)
which ranges between 1 and 1.
These indices can be determined from two capillary pressure
curves corresponding to waterflood and subsequent oilflood, and
can be computed directly from pore-network simulations (ren
and Bakke 2003; Valvatne and Blunt 2004). The variability of the
Amott oil and water indices and the Amott-Harvey wettability
index with respect to the intrinsic contact angle is given in Figs. 18
and 19, respectively. Although this was not pursued here, one
could express the dependence of the trapping and waterflood rela-
tive permeability parameters with respect to the Amott-Harvey
wettability index directly, rather than the intrinsic contact angle.
Conclusions
We have presented a new model of trapping and waterflood rela-
tive permeability. Development of the model is motivated by the
inability of existing models to capture the trends observed for
intermediate-wet and oil-wet media. Because of scarcity of reliable
experimental data, we have used pore-network simulation as a
means to predict the trends in trapping and relative permeability
hysteresis. The new model is able to capture two key features of
the observed behavior: (1) non-monotonicity of the IR curves,
which implies that waterflood relative permeabilities cross; and (2)
convexity of the waterflood relative permeability curves for oil-
wet media caused by layer flow of oil.
The developments presented here are restricted to two-phase
flow. We plan to extend the formulation for trapping and water-
flood relative permeability for three phase flow and, in particular,
for waterflood after gas injection (Spiteri and Juanes 2006).
Nomenclature
b exponent parameter in Jeraulds trapping model
C Land trapping coefficient
I
o
Amott oil wettability index
I
w
Amott water wettability index
I
wo
Amott-Harvey wettability index
k
rw
water relative permeability
k
ro
d
drainage oil relative permeability
k
ro
i
waterflood oil relative permeability
S
o
oil saturation
S
o,max
maximum oil saturation
S
of
flowing oil saturation
S
oi
initial oil saturation
S
ot
ultimate trapped oil saturation
S
w
water saturation
S
o
intermediate trapped oil saturation
initial slope of IR curve in new trapping model
curvature of the IR curve in new trapping model
layer flow parameter in new waterflood model
,
i
intrinsic contact angle

a
advancing contact angle

r
receding contact angle
Acknowledgments
Spiteri, Juanes, and Orr gratefully acknowledge financial support
from the industrial affiliates of the Stanford University Petroleum
Research Institute for Gas Injection (SUPRI-C). Blunt would like
to thank the sponsors of the Imperial College Consortium on Pore-
Scale Modeling and NERC for funding.
References
Al-Futaisi, A. and Patzek, T.W. 2003. Impact of wettability on two-phase
flow characteristics of sedimentary rock: A quasi-static description.
Wat er Resources Research 39 (2): 1042. DOI: 10. 1029/
2002WR001366.
Al-Futaisi, A. and Patzek, T.W. 2004. Secondary imbibition in NAPL-
invaded mixed-wet sediments. J. Contaminant Hydrol. 74 (14): 61
81. DOI:10.1016/j.jconhyd.2004.02.005.
Amott, E. 1959. Observations Relating to the Wettability of Porous Rock.
Trans., AIME, 216: 156162.
Bakke, S. and ren, P-E. 1997. 3-D Pore-Scale Modelling of Sandstones
and Flow Simulations in the Pore Networks. SPEJ 2 (2): 136149.
SPE-35479-PA. DOI: 10.2118/35479-PA.
Blunt, M.J. 2000. An Empirical Model for Three-Phase Relative Perme-
ability. SPEJ 5 (4): 435445. SPE-67950-PA. DOI: 10.2118/67950-
PA.
Blunt, M.J. 2001. Flow in porous mediaPore network models and mul-
tiphase flow. Curr. Opin. Colloid Interface Sci. 6 (3): 197207.
DOI:10.1016/S1359-0294(01)00084-X.
Blunt, M.J. and King, P. 1991. Relative permeabilities from two- and
three-dimensional pore-scale network modeling. Transport in Porous
Media 6 (4): 407433. DOI:10.1007/BF00136349.
Blunt, M.J., Jackson, M.D., Piri, M., and Valvatne P.H. 2003. Detailed
physics, predictive capabilities and macroscopic consequences for
pore-network models of multiphase flow. Advances in Water Resources
25 (812): 10691089. DOI:10.1016/S0309-1708(02)00049-0.
Bryant, S. and Blunt, M.J. 1992. Prediction of relative permeability in
simple porous media. Phys. Rev. A 46 (4): 20042011. DOI:10.1103/
PhysRevA.46.2004.
Carlson, F.M. 1981. Simulation of Relative Permeability Hysteresis to the
Nonwetting Phase. Paper SPE 10157 presented at the SPE Annual
Fig. 15Dependence of parameter of the proposed waterflood
relative permeability model on the intrinsic contact angle of
initially oil-filled pores for a network model of a Berea sandstone.
285 September 2008 SPE Journal
Technical Conference and Exhibition, San Antonio, Texas, USA, 47
October. DOI: 10.2118/10157-MS.
Chatzis, I., Morrow, N.R., and Lim, H.T. 1983. Magnitude and Detailed
Structure of Residual Oil Saturation. SPEJ 23 (2): 311330. SPE-
10681-PA. DOI: 10.2118/10681-PA.
Dardaganian, S.G. 1957. The displacement of gas by oil in the presence of
connate water. MS thesis, Texas A&M University, College Station,
Texas.
Dixit, A.B., McDougall, S.R., Sorbie, K.S., and Buckley, J.S. 1999. Pore-
Scale Modeling of Wettability Effects and Their Influence on Oil Re-
covery. SPEREE 2 (1): 2536. SPE-54454-PA. DOI: 10.2118/54454-
PA.
Donaldson, E.C., Thomas, R.D., and Lorenz, P.B. 1969. Wettability De-
termination and its Effect on Recovery Efficiency. SPEJ 9 (1): 1320;
Trans., AIME, 246. SPE-2338-PA. DOI: 10.2118/2338-PA.
Egermann, P., Vizika, O., Dallet, L., Requin, C., and Sonier, F. Hysteresis
in Three-Phase Flow: Experiments, Modeling and Reservoir Simula-
tions. Paper SPE 65127 presented at the SPE European Petroleum
Conference, Paris, 2425 October. DOI: 10.2118/65127-MS.
Element, D.J., Masters, J.H.K., Sargent, N.C., Jayasekera, A.J., and Good-
year, S.G. 2003. Assesment of Three-Phase Relative Permeability
Models Using Laboratory Hysteresis Data. Paper SPE 84903 presented
at the SPE International Improved Oil Recovery Conference in Asia
Pacific, Kuala Lumpur, 2021 October. DOI: 10.2118/84903-MS.
Fatt, I. 1956. The Network Model of Porous Media. Trans., AIME, 207:
144159.
Geffen, T.M., Owens, W.W., Parrish, D.R., and Morse, R.A. 1951. Ex-
perimental Investigation of Factors Affecting Laboratory Relative Per-
meability Measurements. Trans., AIME, 192: 99110.
Holmgren, C.R. and Morse R.A. 1951. Effect of Free Gas Saturation on Oil
Recovery by Water Flooding. Trans., AIME, 192: 135140.
Holt, R.M., Fjaer, E., Torsaeter, O., and Bakke, S. 1996. Petrophysical
laboratory measurements for basin and reservoir evaluation. Marine
and Petroleum Geology 13 (4): 383391. DOI:10.1016/0264-
8172(95)00091-7.
Ioannidis, M.A. and Chatzis, I. 2000. On the geometry and topology of 3D
stochastic porous media. J. Colloid Interface Sci. 229 (2): 323334.
DOI:10.1006/jcis.2000.7055.
Jackson, M.D., Valvatne, P.H., and Blunt, M.J. 2003. Prediction of wet-
tability variation and its impact on flow using pore- to reservoir-scale
simulations. J. Pet. Sci. Eng. 39 (34): 231246. DOI:10.1016/S0920-
4105(03)00065-2.
Jadhunandan, P.P. and Morrow, N.R. 1995. Effect of Wettability on Wa-
terflood Recovery for Crude-Oil/Brine/Rock Systems. SPERE 10 (1):
4046. SPE-22597-PA. DOI: 10.2118/22597-PA.
Jerauld, G.R. 1997a. General Three-Phase Relative Permeability Model for
Prudhoe Bay. SPERE 12 (4): 255263. SPE-36178-PA. DOI: 10.2118/
36178-PA.
Jerauld, G.R. 1997b. Prudhoe Bay Gas/Oil Relative Permeability. SPERE
12 (1): 6673. SPE-35718-PA. DOI: 10.2118/35718-PA.
Kennedy, H.T., Burja, E.O., and Boykin, R.S. 1955. An investigation of the
effects of wettability on the recovery of oil by water flooding. J. Physi-
cal Chem. 59 (9): 867869. DOI:10.1021/j150531a015.
Killough, J.E. 1976. Reservoir Simulation With History-Dependent Satu-
ration Functions. SPEJ 16 (1): 3748; Trans., AIME, 261. SPE-5106-
PA. DOI: 10.2118/5106-PA.
Kossack, C.A. 2000. Comparison of Reservoir Simulation Hysteresis Op-
tions. Paper SPE 63147 presented at the SPE Annual Technical Con-
ference and Exhibition, Dallas, 14 October. DOI: 10.2118/63147-MS.
Fig. 16Comparison of the new waterflood model with pore-network simulation data for the bounding relative permeability curves
(initial saturation S
oi
=0.76).
286 September 2008 SPE Journal
Kovscek, A.R., Wong, H., and Radke, C.J. 1993. A pore-level scenario for
the development of mixed wettability in oil reservoirs. AIChE Journal
39 (6): 10721085. DOI:10.1002/aic.690390616.
Kyte, J.R., Stanclift, R.J. Jr., Stephan, S.C. Jr., and Rapoport, L.A. 1956.
Mechanism of Water Flooding in the Presence of Free Gas. Trans.,
AIME, 207: 215221.
Land, C.S. 1968. Calculation of Imbibition Relative Permeability for Two-
and Three-Phase Flow From Rock Properties. SPEJ 8 (2): 149156;
Trans., AIME, 243. SPE-1942-PA. DOI: 10.2118/1942-PA.
Land, C.S. 1971. Comparison of Calculated With Experimental Imbibition
Relative Permeability. SPEJ 11 (4): 419425; Trans., AIME, 251. SPE-
3360-PA. DOI: 10.2118/3360-PA.
Fig. 18Amott oil and water wettability indices for all intrinsic
contact angles.
Fig. 19Amott-Harvey wettability index for all intrinsic con-
tact angles.
Fig. 17Comparison of the new waterflood model with pore-network simulation data for the scanning relative permeability curves
(initial saturation S
oi
=0.65).
287 September 2008 SPE Journal
Larsen, J.A. and Skauge, A. 1998. Methodology for numerical simulation
with cycle-dependent relative permeabilities. SPEJ 3 (2): 163173.
SPE-38456-PA. DOI: 10.2118/38456-PA.
Lenhard, R.J. and Oostrom, M. 1998. A parametric model for predicting
relative permeability-saturation-capillary pressure relationships of oil-
water systems in porous media with mixed wettability. Transport in
Porous Media 31 (1): 109131. DOI:10.1023/A:1006503406056.
Lenhard, R.J. and Parker, J.C. 1987. A model for hysteretic constitutive
relations governing multiphase flow 2. Permeability-saturation rela-
tions. Water Resources Research 23 (12): 21972206. DOI:10.1029/
WR023i012p02197.
Lenormand, R., Zarcone, C., and Sarr, A. 1983. Mechanisms of the dis-
placement of one fluid by another in a network of capillary ducts. J.
Fluid Mech. 135: 337353. DOI:10.1017/S0022112083003110.
Moore, T.F. and Slobod, R.L. 1956. The effect of viscosity and capillarity
on the displacement of oil by water. Prod. Monthly 20 (10): 2030.
Morrow, N.R. 1975. The effects of surface roughness on contact angle with
special reference to petroleum recovery. J. Can. Pet. Tech. 14 (Octo-
berDecember): 4254.
Moulu, J-C., Vizika, O., Egermann, P., and Kalaydjian, F. 1999. A New
Three-Phase Relative Permeability Model for Various Wettability Con-
ditions. Paper SPE 56477 presented at the SPE Annual Technical Con-
ference and Exhibition, Houston, 36 October. DOI: 10.2118/56477-
MS.
Oak, M.J. 1990. Three-Phase Relative Permeability of Water-Wet Berea.
Paper SPE 20183 presented at the SPE/DOE Enhanced Oil Recovery
Symposium, Tulsa, 2225 April. DOI: 10.2118/20183-MS.
Oak, M.J. 1991. Three-Phase Relative Permeability of Intermediate-Wet
Berea Sandstone. Paper SPE 22599 presented at the SPE Annual Tech-
nical Conference and Exhibition, Dallas, 69 October. DOI: 10.2118/
22599-MS.
ren, P-E. and Bakke, S. 2002. Process based reconstruction of sandstones
and prediction of transport properties. Transport in Porous Media 46
(23): 311343. DOI:10.1023/A:1015031122338.
ren, P-E. and Bakke, S. 2003. Reconstruction of Berea sandstone and
pore-scale modelling of wettability effects. J. Pet. Sci. Eng. 39 (34):
177199. DOI: 10.1016/S0920-4105(03)00062-7.
ren, P-E., Bakke, S., and Arntzen, O.J. 1998. Extending Predictive Ca-
pabilities to Network Models. SPEJ 3 (4): 324336. SPE-52052-PA.
DOI: 10.2118/52052-PA.
Owens, W.W. and Archer, D.L. 1971. The Effect of Rock Wettability on
Oil-Water Relative Permeability Relationships. JPT 23 (7): 873878.
SPE-3034-PA. DOI: 10.2118/3034-PA.
Patzek T.W. 2001. Verification of a Complete Pore Network Simulator of
Drainage and Imbibition. SPEJ 6 (2): 144156. SPE-71310-PA. DOI:
10.2118/71310-PA.
Spanne, P., Thovert, J.F., Jacquin, C.J., Lindquist, W.B., Jones, K.W., and
Adler, P.M. 1994. Synchrotron computed microtomography of porous
media: topology and transports. Phys. Rev. Lett. 73 (14): 20012004.
DOI: 10.1103/PhysRevLett.73.2001.
Spiteri, E.J. and Juanes, R. 2006. Impact of relative permeability hysteresis
on the numerical simulation of WAG injection. J. Pet. Sci. Eng. 50 (2):
115139. DOI:10.1016/j.petrol.2005.09.004.
Valvatne, P.H. and Blunt, M.J. 2004. Predictive pore-scale modeling of
two-phase flow in mixed wet media. Water Resources Research 40:
W07406. DOI:10.1029/2003WR002627.
van Dijke, M.I.J., Lago, M., Sorbie, K.S., and Araujo M. 2004. Free energy
balance for three fluid phases in a capillary of arbitrarily shaped cross-
section: Capillary entry pressures and layers of the intermediate-
wetting phase. J. Colloid Interface Sci. 277 (1): 184201.
DOI:10.1016/j.jcis.2004.05.021.
Elizabeth Spiteri is a Petroleum Reservoir Engineer at Chevron
Energy Technology Company. e-mail: espiteri@chevron.com.
Her current work focuses on performing simulation studies for
reservoir appraisal and development, currently in West Africa
basins of Nigeria and Angola. Her research interests include
CO
2
sequestration, upscaling and multiscale methods, and 4D
seismic inversion problems. She holds a BS degree in civil and
environmental engineering from the University of Michigan,
and an MS degree in petroleum engineering from Stanford
University. She serves as a technical editor for SPE Reservoir
Evaluation & Engineering Journal. Ruben Juanes is the ARCO
assistant professor in Energy Studies, Department of Civil and
Environmental Engineering, Massachusetts Institute of Technol-
ogy. e-mail: juanes@mit.edu. Before joining the MIT faculty in
2006, he was acting assistant professor in petroleum engineer-
ing at Stanford University, where he was Elizabeth Spiteris re-
search advisor. His interest is in the theory and simulation of
multiphase flow through porous media, with application to en-
hanced oil recovery by gas injection, geological CO
2
seques-
tration, and methane hydrates in nature. He holds MS and PhD
degrees in civil and environmental engineering from the Uni-
versity of California at Berkeley. Martin J. Blunt is Professor of
Petroleum Engineering and head of the Department of Earth
Science and Engineering at Imperial College, London. e-mail:
m.blunt@imperial.ac.uk. He previously was an associate pro-
fessor at Stanford University and worked at the BP Research
Centre. He holds MA and PhD degrees in physics from Cam-
bridge University. Blunt, winner of the 1996 Cedric K. Ferguson
Medal, served as Associate Executive Editor of SPEJ from 1996
98 and was on the Editorial Board from 1996-2005. He is a mem-
ber of the London Board of the SPE. Blunt was a 2001 Distin-
guished Lecturer. Franklin M. Orr, Jr. is the Keleen and Carlton
Beal Professor in Petroleum Engineering, Department of Energy
Resources Engineering, and Director of the Global Climate
and Ener gy Pr oj ect at St anf or d Uni ver si t y. e- mai l :
fmorr@stanford.edu. He served as Dean, School of Earth Sci-
ences, Stanford University from 1994-2002. Previously he held
the positions of head of the miscible flooding section at the
New Mexico Petroleum Recovery Research Center; Adjunct
Associate Professor of Petroleum Engineering at New Mexico
Institute of Mining and Technology; and research engineer at
the Shell Development Company, Bellaire Research Center.
He holds a BS degree from Stanford University and a PhD de-
gree from the University of Minnesota, both in chemical engi-
neering.
288 September 2008 SPE Journal